DNAFix

DNAFix is an autonomous software engineer that fixes bugs the way a careful human engineer would. You give it a GitHub issue or a local repository and it does the rest. It reads the code, works out what will break, writes a plan, makes the change in a safe copy of the repo, runs the tests, has the result checked by a second model, and then asks you before it opens a draft pull request. It never merges anything on its own. Most coding agents just throw a model at the whole repo and hope for the best. DNAFix first runs Project DNA, a deterministic analysis layer that maps the real structure of the code: files, imports, the call graph and the impact of a change. That evidence goes to the model, so the plan is based on what the code actually does instead of a guess. Coding happens in an isolated worktree, so your own working tree is never touched. Tests, type checks and lint decide if the change is good. The model only reviews the evidence. It does not get to claim a test passed when it did not. The problem it solves is trust. Teams want speed, but nobody wants a bot pushing broken code or merging without review. DNAFix keeps a human in the loop at the two points that matter, plan approval and pull request approval, and shows every step in one workspace.